This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

xuru.org

xuru@xuru.org

July 31, 2006

Abstract

Our understanding of Nature relies on calculus, which in turn relies

on the intuitive concept of the derivative. It’s descriptive power comes

from the fact that it analyses the behavior at scales small enough that

its properties change linearly, so avoiding complexities that arise at

larger ones. Fractional Calculus generalizes this concept from integer to

non-integer order. Despite it seems not to have signiﬁcant applications

in fundamental physics, research on this core concept could be valuable

in understanding Nature. These notes comprise an introduction to the

ﬁeld.

1 Introduction

Fractional Calculus is the branch of calculus that generalizes the derivative

of a function to non-integer order, allowing calculations such as deriving a

function to 1/2 order. Despite “generalized” would be a better option, the

name “fractional” is used for denoting this kind of derivative.

!Note: These notes are not a summary of standard methods or an

academic text, so some conventional developments are omitted while

some other are introduced. The same happens with notation.

The derivative of a function f is deﬁned as

D

1

f(x) = lim

h→0

f(x + h) −f(x)

h

(1.1)

Iterating this operation yields an expression for the n-st derivative of a

function. As can be easily seen -and proved by induction- for any natural

number n,

D

n

f(x) = lim

h→0

h

−n

n

¸

m=0

(−1)

m

n

m

f(x + (n −m)h) (1.2)

2 1 Introduction

where

n

m

=

n!

m! (n −m)!

(1.3)

or equivalently,

D

n

f(x) = lim

h→0

h

−n

n

¸

m=0

(−1)

m

n

m

f(x −mh) (1.4)

The case of n = 0 can be included as well.

Such an expression could be valuable for instance in a simple program

for plotting the n-st derivative of a function.

Viewing this expression one asks immediately if it can be generalized to

any non-integer, real or complex number n. There are some reasons that

can make us think so,

1. The fact that for any natural number n the calculation of the n-st

derivative is given by an explicit formula (1.2) or (1.4).

2. That the generalization of the factorial by the gamma function allows

n

m

=

n!

m! (n −m)!

=

Γ(n + 1)

Γ(m + 1)Γ(n −m + 1)

(1.5)

which also is valid for non-integer values.

3. The likeness of (1.2) to the binomial formula

(a + b)

n

=

n

¸

m=0

n

m

a

n−m

b

m

(1.6)

which can be generalized to any complex number α by

(a + b)

α

=

∞

¸

n=0

Γ(α + 1)

n! Γ(α −n + 1)

a

α−n

b

n

(1.7)

which is convergent if

| b | < a (1.8)

There are some desirable properties that could be required to the frac-

tional derivative,

1. Existance and continuity for m times derivable functions, for any n

which modulus is equal or less than m.

2 Exponentials 3

2. For n = 0 the result should be the function itself; for n > 0 integer

values it should be equal to the ordinary derivative and for n < 0

integer values it should be equal to ordinary integration -regardless

the integration constant.

3. Iterating should not give problems,

D

α+β

f(x) = D

α

D

β

f(x) (1.9)

4. Linearity,

D

α

(af(x) + bg(x)) = a D

α

f(x) + b D

α

g(x) (1.10)

5. Allowing Taylor’s expansion in some other way.

6. Its characteristic property should be preserved for the exponential

function,

D

α

e

x

= e

x

(1.11)

2 Exponentials

The case of the exponential function is specially simple and gives some clues

about the generalization of the derivatives. Following (1.2),

D

α

e

ax

= lim

h→0

h

−α

α

¸

n=0

(−1)

n

α

n

e

a(x+(α−n)h)

=

= e

ax

lim

h→0

h

−α

α

¸

n=0

(−1)

n

α

n

(e

ah

)

α−n

=

= e

ax

lim

h→0

h

−α

(e

ah

−1)

α

=

= a

α

e

ax

(2.1)

the above limit exists for any complex number α. However, it should be

noted that in the substitution of the binomial formula a natural number

has been considered. We shall deal with this problem later to get our ﬁrst

generalization of the derivative. Applying this to the imaginary unit,

D

α

cos(x) + i D

α

sin(x) = D

α

e

ix

= i

α

e

ix

=

= e

απi

2

e

ix

= e

i(x+

απ

2

)

=

= cos

x +

απ

2

+ i sin

x +

απ

2

(2.2)

4 2 Exponentials

and

D

α

cos(x) −i D

α

sin(x) = D

α

e

−ix

= (−i)

α

e

−ix

=

= e

−απi

2

e

−ix

= e

−i(x+

απ

2

)

=

= cos

x +

απ

2

−i sin

x +

απ

2

(2.3)

solving this system we have the next deﬁnition for the sine and cosine deriva-

tives,

D

α

sin(x) = sin

x +

απ

2

(2.4)

and

D

α

cos(x) = cos

x +

απ

2

(2.5)

We could expect these relations for the sine and cosine derivatives to be

maintained in the generalization of the derivative.

Applying the above method we also can calculate the following,

D

α

cos(ax) + i D

α

sin(ax) = D

α

e

iax

= (ai)

α

e

iax

=

= a

α

e

απi

2

e

iax

= a

α

e

i(ax+

απ

2

)

=

= a

α

cos

ax +

απ

2

+ ia

α

sin

ax +

απ

2

(2.6)

and

D

α

cos(ax) −i D

α

sin(ax) = D

α

e

−iax

= (−ai)

α

e

−iax

=

= a

α

e

−

απi

2

e

−iax

= a

α

e

−i(ax+

απ

2

)

=

= a

α

cos

ax +

απ

2

−ia

α

sin

ax +

απ

2

(2.7)

Thus,

D

α

sin(ax) = a

α

sin

ax +

απ

2

(2.8)

and

D

α

cos(ax) = a

α

cos

ax +

απ

2

(2.9)

3 Powers 5

Indeed, the above result of the exponential can be applied to any function

that can be expanded in exponentials

f(x) =

∞

¸

n=−∞

a

n

e

inx

=⇒

=⇒ D

α

f(x) =

∞

¸

n=−∞

a

n

D

α

e

inx

=

∞

¸

n=−∞

a

n

(ni)

α

e

inx

=

=

∞

¸

n=−∞

a

n

n

α

e

απi

2

e

inx

=

∞

¸

n=−∞

a

n

n

α

e

i(nx+

απ

2

)

(2.10)

Expanding the function in Fourier series,

f(x) =

∞

¸

n=−∞

e

inx

2π

π

−π

f(t)e

−int

dt =⇒

=⇒ D

α

f(x) =

∞

¸

n=−∞

n

α

e

i(nx+

απ

2

)

2π

π

−π

f(t)e

−int

dt (2.11)

This method can be useful for calculating fractional derivatives of trigono-

metric functions.

3 Powers

The case of powers of x also has some simplicity that allows its generaliza-

tion. The case of integer order derivatives

D

1

x

a

= ax

a−1

=⇒ D

n

x

a

= x

a−n

n−1

¸

m=0

(a −m) =

a!

(a −n)!

x

a−n

(3.1)

can be easily generalized to non-integer order derivatives

D

α

x

a

=

Γ(a + 1)

Γ(a −α + 1)

x

a−α

(3.2)

which can be applied to any function that can be expanded in powers of x

f(x) =

∞

¸

n=−∞

a

n

x

n

=⇒

=⇒ D

α

f(x) =

∞

¸

n=−∞

a

n

D

α

x

n

=

∞

¸

n=−∞

a

n

Γ(n + 1)

Γ(n −α + 1)

x

n−α

(3.3)

6 4 Binomial Formula

Expanding the function in Taylor series,

f(x + a) =

∞

¸

n=0

D

n

f(a)

n!

x

n

=⇒

=⇒ D

α

f(x + a) =

∞

¸

n=0

D

n

f(a)

Γ(n −α + 1)

x

n−α

(3.4)

or expanding it in Laurent series,

f(x + a) =

1

2πi

∞

¸

n=−∞

x

n

C

f(t + a)

t

n+1

dt =⇒

=⇒ D

α

f(x + a) =

1

2πi

∞

¸

n=−∞

Γ(n + 1)

Γ(n −α + 1)

x

n−α

C

f(t + a)

t

n+1

dt

(3.5)

This can be an useful tool for calculating fractional derivatives. How-

ever, we should compare these results of powers with the previous results of

exponentials to see if they agree. With the result of exponentials (2.1),

D

α

e

ax

= a

α

e

ax

= a

α

∞

¸

n=0

(ax)

n

n!

=

∞

¸

n=0

a

n+α

x

n

n!

(3.6)

but with the result of powers (3.4),

D

α

e

ax

= D

α

∞

¸

n=0

(ax)

n

n!

=

∞

¸

n=0

a

n

x

n−α

Γ(n −α + 1)

(3.7)

If we compare both results, we see that they only agree for integer values

of α. We shall see later where these discrepancies come from, and how they

can be avoided.

4 Binomial Formula

In the expression (1.2) the exponentian function allows the substitution of

the binomial formula as done in (2.1), but this is not possible for any given

function. For applying this substitution we require the following displace-

ment operator,

d

h

f(x) = f(x + h) (4.1)

whose iteration yields

d

α

h

f(x) = f(x + αh) (4.2)

5 Functions of the Derivative 7

what allows the application of the binomial formula (1.6) for natural num-

bers and the generalized binomial formula (1.7) for complex numbers,

D

n

f(x) = lim

h→0

h

−n

n

¸

m=0

(−1)

m

n

m

f(x + (n −m)h) =

= lim

h→0

h

−n

n

¸

m=0

(−1)

m

n

m

d

n−m

h

f(x) =

= lim

h→0

d

h

−1

h

n

f(x) (4.3)

so that for any complex number α it can be generalized

D

α

f(x) = lim

h→0

d

h

−1

h

α

f(x) (4.4)

This sheds more light on the derivative and its generalization. Using the

expression of the generalized binomial formula (1.7) for non-integer numbers,

D

α

f(x) = lim

h→0

d

h

−1

h

α

f(x) =

= lim

h→0

h

−α

∞

¸

n=0

Γ(α + 1)

n! Γ(α −n + 1)

(−1)

n

d

α−n

h

f(x) =

= lim

h→0

h

−α

∞

¸

n=0

(−1)

n

Γ(α + 1)

n! Γ(α −n + 1)

f(x + (α −n)h) (4.5)

In the case of integer values the summatory only extends α terms and it

is equal to the ordinary derivative.

Finally, it is obvious that as h goes to 0 the last equation is equivalent

to the following

D

α

f(x) = lim

h→0

h

−α

∞

¸

n=0

(−1)

n

Γ(α + 1)

n! Γ(α −n + 1)

f(x −nh) (4.6)

5 Functions of the Derivative

It is worth to mention that being (4.4) the expression of the α-st derivative

of a function the derivative itself is

D = lim

h→0

d

h

−1

h

(5.1)

And being the α-st derivative or iterating the diﬀerentiation α times

powering it to α, applying other functions to the derivative could be also

8 5 Functions of the Derivative

considered. If the function applied to the derivative can be expanded in

powers of x,

g(x) =

∞

¸

n=−∞

a

n

x

n

=⇒ g(D)f(x) =

∞

¸

n=−∞

a

n

D

n

f(x) (5.2)

The result is a weighted sum of diﬀerent order derivatives. These func-

tions of the derivative are usually known as “formal diﬀerential operators”.

As an example, the exponential of the derivative applied to the exponential

would give the following result that could be valuable for calculating func-

tions of the derivative when both f and g can be expanded in exponentials

e

ax

=

∞

¸

n=0

a

n

x

n

n!

=⇒

=⇒ e

aD

e

bx

=

∞

¸

n=0

a

n

D

n

e

bx

n!

=

∞

¸

n=0

a

n

b

n

e

bx

n!

= e

ab

e

bx

= e

b(x+a)

(5.3)

If both functions f and g can be expanded in positive powers of x,

g(x) =

∞

¸

n=0

a

n

x

n

, f(x) =

∞

¸

n=0

b

n

x

n

=⇒

=⇒ g(D)f(x) =

∞

¸

n=0

a

n

D

n

∞

¸

m=0

b

m

x

m

=

∞

¸

n=0

a

n

∞

¸

m=0

b

m

D

n

x

m

=

=

∞

¸

n=0

a

n

∞

¸

m=0

b

m

Γ(m + 1)

Γ(m−n + 1)

x

m−n

=

=

∞

¸

n=0

a

n

∞

¸

l=−n

b

l+n

Γ(l + n + 1)

Γ(l + 1)

x

l

=

=

∞

¸

n=0

a

n

∞

¸

l=0

b

l+n

(l + n)!

l!

x

l

=

=

∞

¸

l=0

x

l

l!

∞

¸

n=0

(l + n)! a

n

b

l+n

(5.4)

Thus,

g(D)f(x) =

∞

¸

n=0

c

n

x

n

c

n

=

1

n!

∞

¸

m=0

(n + m)! a

m

b

n+m

(5.5)

At a ﬁrst glance this seems not very interesting, but interesting properties

could be hidden under this apparent mess. We shall return later to this point

with the help of other tools.

6 Grunwald-Letnikov Derivative 9

6 Grunwald-Letnikov Derivative

Grunwald-Letnikov derivative or also named Grunwald-Letnikov diﬀerinte-

gral, is a generalization of the derivative analogous to our generalization by

the binomial formula (4.6), but it is based on the direct generalization of the

equation (1.4). The idea behind is that h should approach 0 as n approaches

inﬁnity,

D

n

f(x) = lim

h→0

h

−n

n

¸

m=0

(−1)

m

n

m

f(x −mh) =

= lim

h→0

h

−n

n

¸

m=0

(−1)

m

n!

m! Γ(n −m + 1)

f(x −mh) =

= lim

h→0

h

−n

∞

¸

m=0

(−1)

m

n!

m! Γ(n −m + 1)

f(x −mh) =

= lim

h→0

h

−n

x−a

h

¸

m=0

(−1)

m

n!

m! Γ(n −m + 1)

f(x −mh) (6.1)

being a negative inﬁnity. However, the generalization is done so that any

wished a less than x can be chosen. The above equation can be generalized

now to get a formula equivalent to (4.6)

D

α

f(x) = lim

h→0

h

−α

x−a

h

¸

m=0

(−1)

m

Γ(α + 1)

m! Γ(α −m + 1)

f(x −mh) (6.2)

or equivalently,

D

α

f(x) = lim

n→∞

n

x −a

α

n

¸

m=0

(−1)

m

Γ(α + 1)

m! Γ(α −m + 1)

f

x −m

x −a

n

(6.3)

These generalizations have the inconvenient that if α is a negative integer

they are not valid for the gamma function is inﬁnity in negative integers and

zero. However, for negative α values the following can be used

−n

m

=

−n

¸

l=−n−m+1

l

m!

=

n+m−1

¸

l=n

(−l)

m!

=

(−1)

m

n+m−1

¸

l=n

l

m!

=

=

(−1)

m

(n + m−1)!

m! (n −1)!

=

(−1)

m

Γ(n + m)

m! Γ(n)

(6.4)

so that for negative values equations (6.2) and (6.3) can be changed to

D

−α

f(x) = lim

h→0

h

−α

x−a

h

¸

m=0

Γ(α + m)

m! Γ(α)

f(x −mh) (6.5)

10 7 Riemann-Liouville Derivative

and

D

−α

f(x) = lim

n→∞

n

x −a

α

n

¸

m=0

Γ(α + m)

m! Γ(α)

f

x −m

x −a

n

(6.6)

The reason why x −a is considered in the above equations instead of a

single positive number can be seen considering the case of α = −1, which

must give the same result that integrating the function,

x

a

f(t)dt = D

−1

f(x) = lim

n→∞

h→0

h

n

¸

m=0

Γ(m + 1)

m! Γ(1)

f(x −mh) =

= lim

n→∞

h→0

n

¸

m=0

hf(x −mh) = lim

n→∞

h→0

nh

0

f(x −t)dt =

= lim

n→∞

h→0

x

x−nh

f(t)dt =⇒

=⇒ a = x −nh =⇒ n =

x −a

h

, h =

x −a

n

(6.7)

Yes, in fractional derivatives limits must be considered as they must be in

integrals. Limits only vanish with integer order derivatives. This also means

that fractional derivatives are nonlocal, which may be the reason that makes

this kind of derivatives less useful in describing Nature. We shall see later

more about the integration -or better say diﬀerentiation- limits.

In the case of the derivative generalized by the binomial formula (4.6),

since n goes to inﬁnity regardless of h, x − a must be inﬁnity, so that the

derivative deﬁned in (4.4) and (4.6) is equivalent to the Grunwald-Letnikov

derivative with a lower limit of negative inﬁnity.

7 Riemann-Liouville Derivative

Riemann-Liouville derivative is the most used generalization of the deriva-

tive. It is based on Cauchy’s formula for calculating iterated integrals. If

the ﬁrst integral of a function, which must equal to deriving it to −1, is as

follows

D

−1

f(x) =

x

0

f(t)dt (7.1)

7 Riemann-Liouville Derivative 11

the calculation of the second can be simpliﬁed by interchanging the integra-

tion order

D

−2

f(x) =

x

0

t

2

0

f(t

1

)dt

1

dt

2

=

x

0

x

t

1

f(t

1

)dt

2

dt

1

=

x

0

f(t

1

)

x

t

1

dt

2

dt

1

=

=

x

0

f(t)(x −t)dt (7.2)

This method can be applied repeatedly, resulting in the following formula

for calculating iterated integrals,

D

−n

f(x) =

1

(n −1)!

x

0

f(t)(x −t)

n−1

dt (7.3)

Now this can be easily generalized to non-integer values, in what is the

Riemann-Liouville derivative,

D

α

f(x) =

1

Γ(−α)

x

0

f(t)

(x −t)

α+1

dt (7.4)

Note however that in the above formulas the election of 0 as the lower

limit of integration has been arbitrary, and any other number could be cho-

sen. Generally, the election of the integration limits in this and other gen-

eralizations of the derivative is indicated with subscripts. The Riemann-

Liouville derivative with the lower integration limit a would be

a

D

α

x

f(x) =

1

Γ(−α)

x

a

f(t)

(x −t)

α+1

dt (7.5)

The problem with this generalization is that if the real part of α is

positive or zero the integral diverges. So it only can be used to calculate

generalized integrals. However, this can be solved easily by deriving ﬁrst by

ordinary derivative more than the amount necessary, thus making the re-

maining necessary diﬀerentiation negative and then applying the generalized

derivative for completing the rest in which will be a negative diﬀerentiation,

a

D

α

x

f(x) = D

n

a

D

α−n

x

f(x)

n =

(α)

+ 1 (7.6)

Once seen how integration limits must be speciﬁed in fractional deriva-

tives, we can return to the derivative of the exponential (2.1) and the deriva-

tive of powers (3.2) to see how their disagreement (3.7) was caused by having

12 8 Domain Transforms

diﬀerent limits. With the exponential, supposing b a positive number,

b

−1

e

bx

=

a

D

−1

x

e

bx

=

x

a

e

bx

dx =

e

bx

−e

ba

b

=⇒

e

ba

b

= 0 =⇒

=⇒ a = −∞ (7.7)

while with the powers

x

b+1

b + 1

=

a

D

−1

x

x

b

=

x

a

x

b

dx =

x

b+1

−a

b+1

b + 1

=⇒

a

b+1

b + 1

= 0 =⇒

=⇒ a = 0 (7.8)

So both results were diﬀerent in the case of the exponential function

because the limits of the derivatives were diﬀerent. And the case of this dis-

crepancy is not casual. Diﬀerences in the results of derivatives with diﬀerent

diﬀerentiation limits are indeed important.

Finally, it can be proved that the Grunwald-Letnikov derivative with

any given integration limits is equal to the Riemann-Liouville derivative

with the same limits for any complex number α with a negative real part.

This important result means that our derivative generalized by the binomial

formula in (4.4) and (4.6) is equivalent to the Riemann-Liouville derivative

with a lower limit of negative inﬁnity provided that the real part of α is

negative,

−∞

D

α

x

f(x) = lim

h→0

d

h

−1

h

α

f(x) =

1

Γ(−α)

x

−∞

f(t)

(x −t)

α+1

dt (7.9)

This kind of Riemann-Liouville derivative with a lower limit of negative

inﬁnity is known as the Weyl derivative.

8 Domain Transforms

The Laplace and Fourier transforms that serve to transform to the frequency

domain can be used to get generalizations of the derivative valid for functions

that allow such transformations. The Laplace transform is deﬁned by

L

¸

f(x)

¸

=

∞

0

e

−tx

f(x)dx (8.1)

while its inverse transform is

L

−1

¸

f(x)

¸

=

1

2πi

a+i∞

a−i∞

e

tx

f(x)dx (8.2)

8 Domain Transforms 13

where a is chosen so that it is greather than the real part of any of the

singularities of f(x). An important property of the Laplace transform is

related to the transform of the n-st derivative of a function,

L

¸

D

n

f(x)

¸

= t

n

L

¸

f(x)

¸

−

n−1

¸

m=0

t

m

(D

n−m−1

f)(0) (8.3)

In the case that the terms in the summatory are zero the relation is

particularly simple, and for those kind of functions the derivative can be

generalized so that this property holds true for non-integer values of α

(D

l

0

D

α−m

x

f)(0) = 0 l = 0 . . . m−1 m =

(α)

+ 1 =⇒

=⇒ L

¸

D

α

f(x)

¸

= t

α

L

¸

f(x)

¸

(8.4)

for which the generalized derivative can be deﬁned as

D

α

f(x) = L

−1

t

α

L

¸

f(x)

¸

¸

(8.5)

Keeping in mind the result of the generalized derivative of the exponen-

tial (2.1), the following development provides an easy understanding of the

reasons involved in the above generalization,

D

α

f(x) = D

α

L

−1

L

¸

f(x)

¸

¸

=

= D

α

¸

¸

1

2πi

a+i∞

a−i∞

e

tx

∞

0

e

−tx

f(x)dxdt

¸

=

=

1

2πi

a+i∞

a−i∞

D

α

e

tx

∞

0

e

−tx

f(x)dxdt =

=

1

2πi

a+i∞

a−i∞

t

α

e

tx

∞

0

e

−tx

f(x)dxdt =

= L

−1

t

α

L

¸

f(x)

¸

¸

(8.6)

On the other hand, the Fourier transform is deﬁned by

F

¸

f(x)

¸

=

∞

−∞

e

−itx

f(x)dx (8.7)

while its inverse transform is

F

−1

¸

f(x)

¸

=

1

2π

∞

−∞

e

itx

f(x)dx (8.8)

14 9 Convolution

This transform also has an analogous property related to the transform

of the n-st derivative of a function,

F

¸

D

n

f(x)

¸

= (it)

n

F

¸

f(x)

¸

(8.9)

and the derivative can be generalized so that this property holds true for

non-integer values of α

F

¸

D

α

f(x)

¸

= (it)

α

F

¸

f(x)

¸

(8.10)

yielding the following deﬁnition of the generalized derivative

D

α

f(x) = F

−1

(it)

α

F

¸

f(x)

¸

¸

(8.11)

The application of (2.1) can provide again an easy understanding of the

reasons involved in the generalization,

D

α

f(x) = D

α

F

−1

F

¸

f(x)

¸

¸

=

= D

α

¸

¸

1

2π

∞

−∞

e

itx

∞

−∞

e

−itx

f(x)dxdt

¸

=

=

1

2π

∞

−∞

D

α

e

itx

∞

−∞

e

−itx

f(x)dxdt =

=

1

2π

∞

−∞

(it)

α

e

itx

∞

−∞

e

−itx

f(x)dxdt =

= F

−1

(it)

α

F

¸

f(x)

¸

¸

(8.12)

In these two generalizations the implicit limits of diﬀerentiation should

be determined. In the case of Laplace transform, the generalized derivative

is a Riemann-Liouville derivative with a lower limit of 0, whereas in the

case of Fourier transform it is a Weyl derivative. Indeed, if we check for

instance the derivative generalized by the Fourier transform in the cases of

the sine and cosine functions -calculated in (2.4), (2.5), (2.8) and (2.9) with

the generalized derivative of the exponential that we have seen in (7.7) that

also is a Weyl derivative- we will ﬁnd that they match perfectly.

9 Convolution

The generalizations of the derivative as expressed in (4.6) and (7.4) suggest

that they can be formulated in terms of the convolution, which would be

important for the convolution is a very simple operation in the frequency

9 Convolution 15

spaces achieved by Laplace and Fourier transforms. The following devel-

opment shows how this is the case, and how after all the derivative of a

function is its convolution with certain function,

Φ

α

(x) =

x

α−1

Γ(α)

(9.1)

which Laplace convolution with f(x) yields the Riemann-Liouville derivative

of order −α

Φ

α

(x) ∗ f(x) =

x

0

Φ

α

(x −t)f(t)dt =

x

0

(x −t)

α−1

Γ(−α)

f(t)dt =

= D

−α

f(x) (9.2)

And since the transform of the function expressed in (9.1) gives the

following simple result

L

¸

x

α

¸

=

Γ(α + 1)

t

α+1

=⇒ L{Φ

α

} = L

x

α−1

Γ(α)

¸

= t

−α

(9.3)

and the transform of the convolution is the multiplication of the transforms,

in the case that the function f fulﬁlls all the requirements given previously

for the simplicity of the Laplace transform of the derivative, we again get

the equation (8.4), which was the key for the generalization by Laplace

transform

L

D

−α

f(x)

¸

= L

¸

Φ

α

(x) ∗ f(x)

¸

= L

¸

Φ

α

(x)

¸

L

¸

f(x)

¸

=

= t

−α

L

¸

f(x)

¸

(9.4)

These last results showing that the generalized derivative is a convolution

with certain function, open the interesting question of what other kind of

operators would be deﬁned if functions other than (9.1) had been chosen.

The answer is that the new operators deﬁned would be functions of the

derivative. To see this, we can exploit the linearity of the convolution,

16 9 Convolution

supposing that the function g can be expanded in powers of x

g(x) =

∞

¸

n=−∞

a

n

x

n

=⇒

=⇒ g(D)f(x) =

∞

¸

n=−∞

a

n

D

n

f(x) =

∞

¸

n=−∞

a

n

Φ

−n

(x) ∗ f(x) =

=

¸

∞

¸

n=−∞

a

n

x

−n−1

Γ(−n)

¸

∗ f(x) =

=

¸

∞

¸

n=−∞

a

−n−1

x

n

Γ(n + 1)

¸

∗ f(x) =

= h(x) ∗ f(x) (9.5)

where

h(x) =

∞

¸

n=−∞

b

n

x

n

b

n

=

a

−n−1

Γ(n + 1)

(9.6)

This proves the equivalence between functions of the derivative and con-

volutions. For any convolution there is an equivalent function of the deriva-

tive if the function implied can be expanded in powers of x and vice versa.

Now, the Laplace transform shows what was expectable,

L

¸

g(D)f(x)

¸

= L

¸

h(x) ∗ f(x)

¸

= L

¸

h(x)

¸

L

¸

f(x)

¸

=

= L

∞

¸

n=−∞

a

−n−1

x

n

Γ(n + 1)

L

¸

f(x)

¸

=

=

∞

¸

n=−∞

a

−n−1

L{x

n

}

Γ(n + 1)

L

¸

f(x)

¸

=

=

∞

¸

n=−∞

a

−n−1

t

n+1

L

¸

f(x)

¸

=

=

∞

¸

n=−∞

a

n

t

n

L

¸

f(x)

¸

=

= g(t)L

¸

f(x)

¸

(9.7)

since the linearity of the Laplace transform combined with the result of (8.5)

10 Cauchy Integral Formula 17

shows clearly that

g(D)f(x) =

∞

¸

n=−∞

a

n

D

n

f(x) =

∞

¸

n=−∞

a

n

L

−1

t

n

L

¸

f(x)

¸

¸

=

= L

−1

∞

¸

n=−∞

a

n

t

n

L

¸

f(x)

¸

=

= L

−1

g(t)L

¸

f(x)

¸

¸

(9.8)

as well as (8.11) shows for the Fourier transform that

g(D)f(x) =

∞

¸

n=−∞

a

n

D

n

f(x) =

∞

¸

n=−∞

a

n

F

−1

(it)

n

F

¸

f(x)

¸

¸

=

= F

−1

∞

¸

n=−∞

a

n

(it)

n

F

¸

f(x)

¸

=

= F

−1

g(it)F

¸

f(x)

¸

¸

(9.9)

These are useful tools for the calculation of functions of the derivative.

As an example, the following case is considered with the help of Fourier

transforms

cos(D) sin(x) = F

−1

cos(it)F

¸

sin(x)

¸

¸

=

= F

−1

cos(it)πi

δ(t + 1) −δ(t −1)

¸

=

= cosh(1) sin(x) (9.10)

and directly as in (5.2)

cos(D) sin(x) =

∞

¸

n=0

(−1)

n

(2n)!

D

2n

sin(x) =

∞

¸

n=0

(−1)

n

(2n)!

(−1)

n

sin(x) =

=

∞

¸

n=0

1

(2n)!

sin(x) = cosh(1) sin(x) (9.11)

yielding both methods the same result. The case of (5.3) also matches when

calculated with Fourier transforms.

10 Cauchy Integral Formula

Another way of generalizing the derivative to non-integer order is given by

the Cauchy integral formula that plays a key role in complex analysis,

D

n

f(z) =

n!

2πi

f(t)

(t −z)

n+1

dt (10.1)

18 11 Properties

Despite its generalization to any complex number α seems straightfor-

ward, it must be taken into account that while being n integer there is an

isolated singularity at t = z, being it non-integer there is a branch point,

what means that the integration contour has to be chosen carefully. Other-

wise, the generalization only involves changing the factorial to the gamma

function, so deﬁning what is known as the Cauchy-type fractional derivative

z

0

D

α

z

f(z) =

Γ(α + 1)

2πi

C(z

0

, z

+

)

f(t)

(t −z)

α+1

dt (10.2)

where supposing that the branch line starts at t = z and passes through z

0

,

the contour C starts at t = z

0

, encircles t = z once in the positive sense and

returns to t = z

0

where now the integrand has a diﬀerent value.

It can be proved that this generalization of the derivative is equivalent to

the Riemann-Liouville derivative with a lower limit of z

0

for the appropriate

values of α in which both derivatives are deﬁned.

11 Properties

Fractional derivatives satisfy quite well all the properties that one could

expect from them, despite some of them are only characteristic of integer

order diﬀerentiation and some other have restrictions. For instance, the

property of linearity (1.10) is fulﬁlled, while that of the iteration (1.9) has

some restrictions in the cases that positive diﬀerentiation orders are present.

Some properties that include summatories can be generalized changing

the summatories into integrals. One such property is the expansion in Taylor

series

f(z) =

∞

−∞

a

D

t

z

f(z

0

)

Γ(t + 1)

(z −z

0

)

t

dt (11.1)

and other is the Leibniz rule

a

D

α

z

f(z)g(z) =

∞

−∞

Γ(α + 1)

Γ(t + 1) Γ(α −t + 1)

a

D

α−t

z

f(z)

a

D

t

z

g(z)dt (11.2)

These and other generalized properties can be applied to the study of

special functions, which often can be expressed in terms of simple formu-

las involving fractional derivatives. For instance, Gauss’s hypergeometric

function can be expressed as

2

F

1

(a, b; c; x) =

Γ(a)x

1−c

Γ(b)

0

D

b−c

x

x

b−1

(1 −x)

a

(11.3)

12 Local Operators 19

12 Local Operators

Despite of some interesting results, we have left unanswered important ques-

tions about the functions of the derivative. For instance, it would be very

interesting to know if integer order derivatives are or not the only kind of

functions of the derivative that are local, and if are or not other kind of local

operators that are not functions of the derivative.

Looking at (5.2) it seems that in fact the only local functions of the

derivative are integer order derivatives and their ﬁnite sums, being non-

locality the result of adding inﬁnite terms with displacement functions

steadily increasing the distance to the point in which the function is evalu-

ated. However, this impediment can be overcome by deﬁning the derivative

in the following way

D

α

f(x) = lim

a→x

a

D

α

x

f(x) (12.1)

so that the limit assures locality. In order to avoid another feature of frac-

tional derivatives that sometimes is not wanted -the fact that the derivative

of a constant is not zero- it is usual to deﬁne the local fractional derivative

as

D

α

f(x) = lim

y→x

x

D

α

y

f(y) −f(x)

(12.2)

or similar forms. So there can be operators being local and taking an inﬁnite

number of evaluations of the function, provided that the points in which

these evaluations are carried out -or the value of their displacement function-

remains being inﬁnitesimal. The question now is if there are other local

operators besides the local fractional derivative and functions of it.

It could be argued that if a function admits its expansion in Taylor

series, all the information about the function is indeed in its derivatives, so

that any operator giving information about the function will be unnecessary

and reducible to a sum of integer order derivatives. However, on the one

hand there are functions that are not diﬀerentiable but are local fractional

diﬀerentiable, and on the other hand, even if local operators could be reduced

to ordinary derivatives for diﬀerentiable functions, that does not assure that

ordinary derivatives are the most appropriate local operators for all the

tasks.

Characterizing all local operators is important, for locality makes of them

tools that could be useful in understanding Nature.

Such an expression could be valuable for instance in a simple program for plotting the n-st derivative of a function. 3. 1. real or complex number n.7) which is convergent if |b| < a (1. The fact that for any natural number n the calculation of the n-st derivative is given by an explicit formula (1.3) D n f (x) = lim h−n h→0 m=0 (−1)m n f (x − mh) m (1. The likeness of (1.2 1 Introduction where n m or equivalently. for any n which modulus is equal or less than m. Existance and continuity for m times derivable functions. There are some reasons that can make us think so.5) which also is valid for non-integer values.2) or (1. n = n! m!(n − m)! (1.4) The case of n = 0 can be included as well. Viewing this expression one asks immediately if it can be generalized to any non-integer. .2) to the binomial formula n (a + b) = m=0 n n n−m m a b m (1. 2.6) which can be generalized to any complex number α by ∞ (a + b)α = n=0 Γ(α + 1) aα−n bn n! Γ(α − n + 1) (1.8) There are some desirable properties that could be required to the fractional derivative. That the generalization of the factorial by the gamma function allows n m = n! Γ(n + 1) = m!(n − m)! Γ(m + 1)Γ(n − m + 1) (1. 1.4).

D α+β f (x) = D α D β f (x) 4. 6.2).11) (1. Allowing Taylor’s expansion in some other way.1) the above limit exists for any complex number α. α D α eax = lim h−α h→0 n=0 (−1)n α α a(x+(α−n)h) e = n α (eah )α−n = n = eax lim h−α h→0 (−1)n (e n=0 ah =e ax =a e h→0 α ax lim h −α − 1)α = (2.2) = cos x + + i sin x + .10) (1. For n = 0 the result should be the function itself. D α ex = ex (1. Its characteristic property should be preserved for the exponential function. for n > 0 integer values it should be equal to the ordinary derivative and for n < 0 integer values it should be equal to ordinary integration -regardless the integration constant.9) 2 Exponentials The case of the exponential function is specially simple and gives some clues about the generalization of the derivatives. Following (1. However. Iterating should not give problems.2 Exponentials 3 2. D α (af (x) + bg(x)) = a D α f (x) + b D α g(x) 5. Applying this to the imaginary unit. Linearity. We shall deal with this problem later to get our ﬁrst generalization of the derivative. it should be noted that in the substitution of the binomial formula a natural number has been considered. D α cos(x) + i D α sin(x) = D α eix = iα eix = =e απ i 2 eix = ei(x+ απ 2 απ ) 2 = απ 2 (2. 3.

D α sin(x) = sin x + and D α cos(x) = cos x + απ 2 (2. D α sin(ax) = aα sin ax + and D α cos(ax) = aα cos ax + απ 2 (2. Applying the above method we also can calculate the following.6) = aα cos ax + + iaα sin ax + and D α cos(ax) − i D α sin(ax) = D α e−iax = (−ai)α e−iax = = aα e− απ i 2 e−iax = aα e−i(ax+ απ 2 απ ) 2 = απ 2 (2.8) .7) = aα cos ax + − iaα sin ax + Thus.5) απ 2 (2.4 2 Exponentials and D α cos(x) − i D α sin(x) = D α e−ix = (−i)α e−ix = =e −απ i 2 e−ix = e−i(x+ απ 2 απ ) 2 = απ 2 (2. D α cos(ax) + i D α sin(ax) = D α eiax = (ai)α eiax = = aα e απ i 2 eiax = aα ei(ax+ απ 2 απ ) 2 = απ 2 (2.4) We could expect these relations for the sine and cosine derivatives to be maintained in the generalization of the derivative.9) απ 2 (2.3) = cos x + − i sin x + solving this system we have the next deﬁnition for the sine and cosine derivatives.

3) .1) can be easily generalized to non-integer order derivatives D α xa = Γ(a + 1) xa−α Γ(a − α + 1) (3. 3 Powers The case of powers of x also has some simplicity that allows its generalization.2) which can be applied to any function that can be expanded in powers of x ∞ f (x) = an xn =⇒ ∞ ∞ n=−∞ =⇒ D α f (x) = an D α xn = an n=−∞ n=−∞ Γ(n + 1) xn−α Γ(n − α + 1) (3. the above result of the exponential can be applied to any function that can be expanded in exponentials ∞ f (x) = an einx =⇒ ∞ ∞ n=−∞ =⇒ D α f (x) = = an D α einx = an nα e απ i 2 an (ni)α einx = = an nα ei(nx+ απ ) 2 n=−∞ ∞ n=−∞ e n=−∞ ∞ inx (2.10) n=−∞ Expanding the function in Fourier series. einx f (x) = 2π n=−∞ α ∞ π f (t)e−int dt =⇒ −π ∞ απ ) 2 nα ei(nx+ =⇒ D f (x) = 2π n=−∞ π f (t)e−int dt −π (2. The case of integer order derivatives n−1 D 1 xa = axa−1 =⇒ D n xa = xa−n m=0 (a − m) = a! xa−n (a − n)! (3.11) This method can be useful for calculating fractional derivatives of trigonometric functions.3 Powers 5 Indeed.

and how they can be avoided. we should compare these results of powers with the previous results of exponentials to see if they agree. we see that they only agree for integer values of α.4) or expanding it in Laurent series. However.5) This can be an useful tool for calculating fractional derivatives.6) but with the result of powers (3.1) (4.6 4 Binomial Formula Expanding the function in Taylor series. ∞ D α eax = aα eax = aα n=0 (ax)n = n! ∞ n=0 an+α xn n! (3.7) If we compare both results. ∞ D α eax = D α n=0 (ax)n = n! ∞ n=0 an xn−α Γ(n − α + 1) (3. ∞ f (x + a) = n=0 D n f (a) n x =⇒ n! ∞ =⇒ D α f (x + a) = n=0 D n f (a) xn−α Γ(n − α + 1) (3.2) . f (x + a) = 1 xn 2πi n=−∞ ∞ f (t + a) dt =⇒ tn+1 C =⇒ D α f (x + a) = Γ(n + 1) 1 xn−α 2πi n=−∞ Γ(n − α + 1) ∞ f (t + a) dt tn+1 C (3. For applying this substitution we require the following displacement operator. With the result of exponentials (2. We shall see later where these discrepancies come from.2) the exponentian function allows the substitution of the binomial formula as done in (2.1). 4 Binomial Formula In the expression (1. but this is not possible for any given function.1). d h f (x) = f (x + h) whose iteration yields α d h f (x) = f (x + αh) (4.4).

Using the expression of the generalized binomial formula (1. applying other functions to the derivative could be also .1) h→0 And being the α-st derivative or iterating the diﬀerentiation α times powering it to α.6) 5 Functions of the Derivative It is worth to mention that being (4.5 Functions of the Derivative 7 what allows the application of the binomial formula (1.4) This sheds more light on the derivative and its generalization. Finally. D α f (x) = lim dh − 1 h ∞ α h→0 f (x) = Γ(α + 1) α−n (−1)n d h f (x) = n! Γ(α − n + 1) = lim h−α h→0 n=0 ∞ n=0 = lim h−α h→0 (−1)n Γ(α + 1) f (x + (α − n)h) n! Γ(α − n + 1) (4.6) for natural numbers and the generalized binomial formula (1. n D n f (x) = lim h−n h→0 m=0 n (−1)m (−1)m m=0 n n f (x + (n − m)h) = m n d n−m f (x) = m h (4.3) = lim h−n h→0 = lim h→0 dh − 1 h f (x) so that for any complex number α it can be generalized D α f (x) = lim dh − 1 h α h→0 f (x) (4.7) for non-integer numbers.4) the expression of the α-st derivative of a function the derivative itself is D = lim dh − 1 h (5.5) In the case of integer values the summatory only extends α terms and it is equal to the ordinary derivative.7) for complex numbers. it is obvious that as h goes to 0 the last equation is equivalent to the following ∞ D α f (x) = lim h−α h→0 n=0 (−1)n Γ(α + 1) f (x − nh) n! Γ(α − n + 1) (4.

5) At a ﬁrst glance this seems not very interesting.3) n! If both functions f and g can be expanded in positive powers of x. ∞ g(D)f (x) = n=0 cn xn ∞ 1 cn = n! (n + m)! am bn+m m=0 (5. If the function applied to the derivative can be expanded in powers of x. These functions of the derivative are usually known as “formal diﬀerential operators”. f (x) = n=0 ∞ bn xn =⇒ ∞ ∞ ∞ n m=0 ∞ =⇒ g(D)f (x) = n=0 ∞ an D an n=0 ∞ bm x bm m = n=0 an m=0 bm D n xm = = = n=0 ∞ m=0 ∞ Γ(m + 1) xm−n = Γ(m − n + 1) Γ(l + n + 1) l x = Γ(l + 1) an l=−n ∞ bl+n bl+n l=0 ∞ n=0 = n=0 ∞ an xl l! (l + n)! l x = l! (5. .2) n=−∞ n=−∞ The result is a weighted sum of diﬀerent order derivatives.4) = l=0 (l + n)! an bl+n Thus. As an example. We shall return later to this point with the help of other tools. but interesting properties could be hidden under this apparent mess. ∞ ∞ g(x) = n=0 an x n . ∞ ∞ g(x) = an x n =⇒ g(D)f (x) = an D n f (x) (5.8 5 Functions of the Derivative considered. the exponential of the derivative applied to the exponential would give the following result that could be valuable for calculating functions of the derivative when both f and g can be expanded in exponentials ∞ eax = n=0 an xn =⇒ n! ∞ =⇒ eaD ebx = n=0 an D n ebx = n! ∞ n=0 an bn ebx = eab ebx = eb(x+a) (5.

The above equation can be generalized now to get a formula equivalent to (4.4).5) .3) These generalizations have the inconvenient that if α is a negative integer they are not valid for the gamma function is inﬁnity in negative integers and zero. However.6 Grunwald-Letnikov Derivative 9 6 Grunwald-Letnikov Derivative Grunwald-Letnikov derivative or also named Grunwald-Letnikov diﬀerintegral. for negative α values the following can be used −n n+m−1 l −n m = l=−n−m+1 l=n (−l) n+m−1 m (−1) l l=n = = m! m! m! (−1)m (n + m − 1)! (−1)m Γ(n + m) = = m!(n − 1)! m! Γ(n) = (6. but it is based on the direct generalization of the equation (1. n D f (x) = lim h h→0 n −n m=0 n (−1)m (−1)m m=0 ∞ n f (x − mh) = m n! f (x − mh) = m! Γ(n − m + 1) n! f (x − mh) = m! Γ(n − m + 1) n! f (x − mh) m! Γ(n − m + 1) (6.6) x−a h D f (x) = lim h h→0 α −α m=0 (−1)m Γ(α + 1) f (x − mh) m! Γ(α − m + 1) (6.6).2) or equivalently. The idea behind is that h should approach 0 as n approaches inﬁnity. is a generalization of the derivative analogous to our generalization by the binomial formula (4. D α f (x) = lim n→∞ n x−a α n (−1)m m=0 Γ(α + 1) x−a f x−m m! Γ(α − m + 1) n (6.3) can be changed to x−a h D −α f (x) = lim h h→0 −α m=0 Γ(α + m) f (x − mh) m! Γ(α) (6.2) and (6. the generalization is done so that any wished a less than x can be chosen. However.4) so that for negative values equations (6.1) = lim h−n h→0 = lim h−n h→0 m=0 x−a h (−1)m = lim h−n h→0 m=0 (−1)m being a negative inﬁnity.

1) . in fractional derivatives limits must be considered as they must be in integrals. which may be the reason that makes this kind of derivatives less useful in describing Nature. This also means that fractional derivatives are nonlocal. x − a must be inﬁnity. h= x−a n (6. is as follows x D −1 f (x) = 0 f (t)dt (7. which must equal to deriving it to −1. In the case of the derivative generalized by the binomial formula (4. 7 Riemann-Liouville Derivative Riemann-Liouville derivative is the most used generalization of the derivative.6) is equivalent to the Grunwald-Letnikov derivative with a lower limit of negative inﬁnity. Limits only vanish with integer order derivatives. so that the derivative deﬁned in (4.10 7 Riemann-Liouville Derivative and D −α f (x) = lim n x−a α n m=0 n→∞ Γ(α + m) x−a f x−m m! Γ(α) n (6.6) The reason why x − a is considered in the above equations instead of a single positive number can be seen considering the case of α = −1. If the ﬁrst integral of a function. x n −1 f (t)dt = D a f (x) = n→∞ h lim h→0 m=0 n Γ(m + 1) f (x − mh) = m! Γ(1) nh = n→∞ lim h→0 m=0 x hf (x − mh) = n→∞ lim h→0 0 f (x − t)dt = = n→∞ lim h→0 f (t)dt =⇒ x−nh =⇒ a = x − nh =⇒ n = x−a h .4) and (4. It is based on Cauchy’s formula for calculating iterated integrals.7) Yes. We shall see later more about the integration -or better say diﬀerentiation. which must give the same result that integrating the function. since n goes to inﬁnity regardless of h.limits.6).

thus making the remaining necessary diﬀerentiation negative and then applying the generalized derivative for completing the rest in which will be a negative diﬀerentiation.1) and the derivative of powers (3. resulting in the following formula for calculating iterated integrals.7 Riemann-Liouville Derivative 11 the calculation of the second can be simpliﬁed by interchanging the integration order x t2 x x x x D −2 f (x) = 0 0 x f (t1 )dt1 dt2 = 0 t1 f (t1 )dt2 dt1 = 0 f (t1 ) dt2 dt1 = t1 = 0 f (t)(x − t)dt (7.3) Now this can be easily generalized to non-integer values.7) was caused by having . So it only can be used to calculate generalized integrals. Generally. we can return to the derivative of the exponential (2. in what is the Riemann-Liouville derivative. x 1 D f (x) = Γ(−α) α 0 f (t) dt (x − t)α+1 (7.4) Note however that in the above formulas the election of 0 as the lower limit of integration has been arbitrary. x D −n 1 f (x) = (n − 1)! 0 f (t)(x − t)n−1 dt (7. this can be solved easily by deriving ﬁrst by ordinary derivative more than the amount necessary. The RiemannLiouville derivative with the lower integration limit a would be x α a D x f (x) 1 = Γ(−α) a f (t) dt (x − t)α+1 (7.2) This method can be applied repeatedly. However.5) The problem with this generalization is that if the real part of α is positive or zero the integral diverges.6) Once seen how integration limits must be speciﬁed in fractional derivatives. α a D x f (x) α−n = D n a Dx f (x) n= (α) + 1 (7.2) to see how their disagreement (3. and any other number could be chosen. the election of the integration limits in this and other generalizations of the derivative is indicated with subscripts.

Diﬀerences in the results of derivatives with diﬀerent diﬀerentiation limits are indeed important.7) =⇒ a = −∞ while with the powers xb+1 −1 = a Dx xb = b+1 a x xb dx = xb+1 − ab+1 ab+1 =⇒ = 0 =⇒ b+1 b+1 (7. it can be proved that the Grunwald-Letnikov derivative with any given integration limits is equal to the Riemann-Liouville derivative with the same limits for any complex number α with a negative real part.2) .6) is equivalent to the Riemann-Liouville derivative with a lower limit of negative inﬁnity provided that the real part of α is negative. And the case of this discrepancy is not casual. x b −1 bx e = −1 bx a Dx e = a ebx dx = eba ebx − eba =⇒ = 0 =⇒ b b (7.1) while its inverse transform is a+i∞ L −1 1 f (x) = 2πi etx f (x)dx a−i∞ (8. supposing b a positive number.12 8 Domain Transforms diﬀerent limits. 8 Domain Transforms The Laplace and Fourier transforms that serve to transform to the frequency domain can be used to get generalizations of the derivative valid for functions that allow such transformations.4) and (4. α −∞ D x f (x) = lim h→0 dh − 1 h α x 1 f (x) = Γ(−α) −∞ f (t) dt (x − t)α+1 (7.8) =⇒ a = 0 So both results were diﬀerent in the case of the exponential function because the limits of the derivatives were diﬀerent. This important result means that our derivative generalized by the binomial formula in (4.9) This kind of Riemann-Liouville derivative with a lower limit of negative inﬁnity is known as the Weyl derivative. With the exponential. The Laplace transform is deﬁned by ∞ L f (x) = 0 e−tx f (x)dx (8. Finally.

7) while its inverse transform is ∞ F −1 1 f (x) = 2π −∞ eitx f (x)dx (8.5) Keeping in mind the result of the generalized derivative of the exponential (2. the Fourier transform is deﬁned by ∞ F f (x) = −∞ e−itx f (x)dx (8... the following development provides an easy understanding of the reasons involved in the above generalization.6) = L−1 tα L f (x) On the other hand.4) =⇒ L D f (x) = t L f (x) for which the generalized derivative can be deﬁned as D α f (x) = L−1 tα L f (x) (8.8) . An important property of the Laplace transform is related to the transform of the n-st derivative of a function. and for those kind of functions the derivative can be generalized so that this property holds true for non-integer values of α α−m (D l 0 Dx f )(0) = 0 l = 0. n−1 L D n f (x) = tn L f (x) − m=0 tm (D n−m−1 f )(0) (8. D α f (x) = D α L−1 L f (x) a+i∞ = ∞ 1 = Dα 2πi a+i∞ etx a−i∞ 0 ∞ e−tx f (x)dxdt = = 1 2πi 1 2πi D α etx a−i∞ a+i∞ 0 ∞ e−tx f (x)dxdt = = tα etx a−i∞ 0 e−tx f (x)dxdt = (8.3) In the case that the terms in the summatory are zero the relation is particularly simple.m − 1 α α m= (α) + 1 =⇒ (8.8 Domain Transforms 13 where a is chosen so that it is greather than the real part of any of the singularities of f (x).1).

1) can provide again an easy understanding of the reasons involved in the generalization.6) and (7.12) = F (it) F f (x) In these two generalizations the implicit limits of diﬀerentiation should be determined.11) (8. F D n f (x) = (it)n F f (x) (8.9) with the generalized derivative of the exponential that we have seen in (7. D α f (x) = D α F −1 F f (x) ∞ = e−itx f (x)dxdt = ∞ 1 = Dα 2π −∞ ∞ eitx −∞ ∞ 1 = 2π −∞ ∞ D e α itx e−itx f (x)dxdt = −∞ ∞ = 1 2π −1 (it)α eitx −∞ −∞ α e−itx f (x)dxdt = (8.10) The application of (2. Indeed. whereas in the case of Fourier transform it is a Weyl derivative.5). if we check for instance the derivative generalized by the Fourier transform in the cases of the sine and cosine functions -calculated in (2. 9 Convolution The generalizations of the derivative as expressed in (4. (2.we will ﬁnd that they match perfectly.4).8) and (2. which would be important for the convolution is a very simple operation in the frequency .7) that also is a Weyl derivative.14 9 Convolution This transform also has an analogous property related to the transform of the n-st derivative of a function.9) and the derivative can be generalized so that this property holds true for non-integer values of α F D α f (x) = (it)α F f (x) yielding the following deﬁnition of the generalized derivative D α f (x) = F −1 (it)α F f (x) (8.4) suggest that they can be formulated in terms of the convolution. the generalized derivative is a Riemann-Liouville derivative with a lower limit of 0. In the case of Laplace transform. (2.

To see this. and how after all the derivative of a function is its convolution with certain function.1) had been chosen. in the case that the function f fulﬁlls all the requirements given previously for the simplicity of the Laplace transform of the derivative. .1) which Laplace convolution with f (x) yields the Riemann-Liouville derivative of order −α x x Φα (x) ∗ f (x) = 0 Φα (x − t)f (t)dt = 0 −α (x − t)α−1 f (t)dt = Γ(−α) (9. xα−1 Γ(α) Φα (x) = (9.4) These last results showing that the generalized derivative is a convolution with certain function. we can exploit the linearity of the convolution. which was the key for the generalization by Laplace transform L D −α f (x) = L Φα (x) ∗ f (x) = L Φα (x) L f (x) = = t−α L f (x) (9. we again get the equation (8.9 Convolution 15 spaces achieved by Laplace and Fourier transforms. open the interesting question of what other kind of operators would be deﬁned if functions other than (9.2) = D f (x) And since the transform of the function expressed in (9. The answer is that the new operators deﬁned would be functions of the derivative.3) and the transform of the convolution is the multiplication of the transforms.1) gives the following simple result L xα = Γ(α + 1) =⇒ L {Φα } = L tα+1 xα−1 Γ(α) = t−α (9.4). The following development shows how this is the case.

6) n=−∞ This proves the equivalence between functions of the derivative and convolutions. the Laplace transform shows what was expectable. L g(D)f (x) = L h(x) ∗ f (x) = L h(x) L f (x) = ∞ a n −n−1 x L f (x) = = L Γ(n + 1) n=−∞ = a−n−1 L {xn } L f (x) = Γ(n + 1) n=−∞ a−n−1 L f (x) = tn+1 n=−∞ ∞ ∞ ∞ = = an tn L f (x) = (9.7) n=−∞ = g(t)L f (x) since the linearity of the Laplace transform combined with the result of (8.5) .5) h(x) = bn = bn xn a−n−1 Γ(n + 1) (9. For any convolution there is an equivalent function of the derivative if the function implied can be expanded in powers of x and vice versa. Now.16 9 Convolution supposing that the function g can be expanded in powers of x ∞ g(x) = an xn =⇒ ∞ ∞ n=−∞ =⇒ g(D)f (x) = = an D f (x) = ∞ n an Φ−n (x) ∗ f (x) = n=−∞ n=−∞ an ∗ f (x) = Γ(−n) n=−∞ ∞ n a−n−1 x ∗ f (x) = = Γ(n + 1) n=−∞ = h(x) ∗ f (x) where ∞ x−n−1 (9.

1) .10) = F −1 cos(it)πi δ(t + 1) − δ(t − 1) = cosh(1) sin(x) and directly as in (5.2) ∞ cos(D) sin(x) = n=0 ∞ (−1)n 2n D sin(x) = (2n)! ∞ n=0 (−1)n (−1)n sin(x) = (2n)! (9.11) shows for the Fourier transform that ∞ ∞ g(D)f (x) = an D n f (x) = ∞ an F −1 (it)n F f (x) = n=−∞ n=−∞ = F −1 an (it)n F f (x) = (9.11) = n=0 1 sin(x) = cosh(1) sin(x) (2n)! yielding both methods the same result. 10 Cauchy Integral Formula Another way of generalizing the derivative to non-integer order is given by the Cauchy integral formula that plays a key role in complex analysis.9) n=−∞ = F −1 g(it)F f (x) These are useful tools for the calculation of functions of the derivative. D n f (z) = n! 2πi f (t) dt (t − z)n+1 (10. As an example.8) n=−∞ = L−1 g(t)L f (x) as well as (8.10 Cauchy Integral Formula 17 shows clearly that ∞ ∞ g(D)f (x) = an D n f (x) = ∞ an L−1 tn L f (x) = n=−∞ n=−∞ = L−1 an tn L f (x) = (9. The case of (5.3) also matches when calculated with Fourier transforms. the following case is considered with the help of Fourier transforms cos(D) sin(x) = F −1 cos(it)F sin(x) = = (9.

It can be proved that this generalization of the derivative is equivalent to the Riemann-Liouville derivative with a lower limit of z0 for the appropriate values of α in which both derivatives are deﬁned. despite some of them are only characteristic of integer order diﬀerentiation and some other have restrictions. being it non-integer there is a branch point.2) These and other generalized properties can be applied to the study of special functions. f (t) dt (t − z)α+1 z+ ) (10. Some properties that include summatories can be generalized changing the summatories into integrals. what means that the integration contour has to be chosen carefully. For instance. For instance. the generalization only involves changing the factorial to the gamma function. which often can be expressed in terms of simple formulas involving fractional derivatives. encircles t = z once in the positive sense and returns to t = z0 where now the integrand has a diﬀerent value.1) and other is the Leibniz rule ∞ α a D z f (z)g(z) = −∞ Γ(α + 1) t D α−t f (z) a Dz g(z)dt Γ(t + 1) Γ(α − t + 1) a z (11. Otherwise.18 11 Properties Despite its generalization to any complex number α seems straightforward. so deﬁning what is known as the Cauchy-type fractional derivative α z0 D z f (z) = Γ(α + 1) 2πi C (z0 . One such property is the expansion in Taylor series ∞ f (z) = −∞ t a Dz f (z0 ) (z − z0 )t dt Γ(t + 1) (11. while that of the iteration (1. b.3) . the property of linearity (1. 11 Properties Fractional derivatives satisfy quite well all the properties that one could expect from them. Gauss’s hypergeometric function can be expressed as 2 F 1 (a. x) = xb−1 Γ(a)x1−c b−c 0 Dx Γ(b) (1 − x)a (11.9) has some restrictions in the cases that positive diﬀerentiation orders are present. c. it must be taken into account that while being n integer there is an isolated singularity at t = z.2) where supposing that the branch line starts at t = z and passes through z0 .10) is fulﬁlled. the contour C starts at t = z0 .

this impediment can be overcome by deﬁning the derivative in the following way D α f (x) = lim D α f (x) a→x a x (12.2) or similar forms. it would be very interesting to know if integer order derivatives are or not the only kind of functions of the derivative that are local. being nonlocality the result of adding inﬁnite terms with displacement functions steadily increasing the distance to the point in which the function is evaluated. Characterizing all local operators is important. and if are or not other kind of local operators that are not functions of the derivative. for locality makes of them tools that could be useful in understanding Nature.1) so that the limit assures locality.it is usual to deﬁne the local fractional derivative as D α f (x) = lim Dα y→x x y f (y) − f (x) (12. For instance. all the information about the function is indeed in its derivatives. provided that the points in which these evaluations are carried out -or the value of their displacement functionremains being inﬁnitesimal. Looking at (5. and on the other hand. so that any operator giving information about the function will be unnecessary and reducible to a sum of integer order derivatives. However. on the one hand there are functions that are not diﬀerentiable but are local fractional diﬀerentiable. In order to avoid another feature of fractional derivatives that sometimes is not wanted -the fact that the derivative of a constant is not zero. we have left unanswered important questions about the functions of the derivative. It could be argued that if a function admits its expansion in Taylor series. . that does not assure that ordinary derivatives are the most appropriate local operators for all the tasks.2) it seems that in fact the only local functions of the derivative are integer order derivatives and their ﬁnite sums.12 Local Operators 19 12 Local Operators Despite of some interesting results. even if local operators could be reduced to ordinary derivatives for diﬀerentiable functions. The question now is if there are other local operators besides the local fractional derivative and functions of it. So there can be operators being local and taking an inﬁnite number of evaluations of the function. However.

- Div
- Real Penting
- Analisis Matmático
- Mth404 Notes
- Chapter 8
- Definite Integral Intro
- partial derivatives
- alg_notes_2_14
- De Summary 1
- 1-s2.0-0022247X61900518-main
- Chapter 3_Power Series
- Chapter 020
- Partial Differentiatiion
- Chapter 12 MATB 113
- Chapter 12 MATB 113
- Rational and Irrational Numbers
- Real Analysis:)
- IB Mathematical Methods II Part 5 of 6 (Cambridge)
- Sequences and Series of Functions
- All_Complex_Analysis_Lectures.pdf
- abstract algebra 2
- MA1506CHAP4
- Algebra 2 Crash Course Semester 1
- Metric and Topological Spaces notes
- IBHM_183-216
- Vectors - 3.1, 3.2, 3.3, 3.4, 3.5
- RudinCh7 Notes
- jumarieAMS37-40-2008[1]

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd